Update README.md

README.md

@@ -1,27 +1,3 @@
-# HuggingFace Space: Computer UI Understanding (CUI)
-
-This Space automates the process of extracting frames from videos, tracking cursor actions, and generating structured annotations for UI understanding datasets.
-
-## Directory Structure
-- `videos/` — Place your input videos here.
-- `frames/` — Extracted frames will be saved here.
-- `cursors/` — Cursor templates for detection.
-- `annotations/` — Output JSON annotation files.
-- `scripts/` — All processing scripts (frame extraction, cursor tracking, annotation, API, etc).
-
-## Usage
-1. Upload your video(s) to the `videos/` directory.
-2. Run the pipeline (see below) to extract frames, track cursor, and generate annotations.
-3. All outputs will be saved in the appropriate folders.
-
-## Automation
-The pipeline is orchestrated by the `scripts/pipeline.py` script, which runs all steps in order.
-
-## HuggingFace Space Notes
-- The Space is fully writeable; all outputs are saved in the workspace.
-- The Docker container is configured for all dependencies and write permissions.
-
-## API
-A FastAPI server is provided for vision model inference.
-
----
+---
+colorFrom: yellow
+---